Investigation of aircraft auxiliary power unit acoustic signatures for condition monitoring

Abstract

The auxiliary power unit (APU) of an aircraft is a key system responsible for providing electrical and pneumatic power during ground operations and in-flight emergencies. APU failures can result in delay or cancellation of a flight and fault diagnostic practices are in place to identify the cause of failure. The existing strategies generally require human intervention to identify the fault by traversing through a troubleshooting manual and examining the data acquired from multiple intrusive sensors. The complete process is cumbersome and prone to misjudgement; fault identification in its entirety may not be possible due to limited sensor coverage. Incorporating additional sensors may not be feasible due to accessibility issues, space constraints and certification requirements. However, incorporating microphones, which have previously been used for noise source characterization and verification of noise abatement solutions, is a promising non- intrusive approach. This PhD focuses on ascertaining the potential of microphones for fault detection / identification and condition monitoring of an aircraft APU. The research has been based on the far-field and near-field acoustic data acquired from Cranfield University’s Boeing 737-400 aircraft and the aim has been to determine the degradation / faults that can be detected using microphones. While addressing this aim, a far-field noise model has been developed for sensitivity analysis, near-field data has been analysed, classification / regression models have been proposed and an acoustics-based scheme for ignition system monitoring has been conceived. The results suggest that the far-field acoustic data is not suitable for condition monitoring, and the near-field microphones are unable to monitor tonal frequencies for monitoring the gearbox and bearing. However, there is a huge potential in using microphones for monitoring the lubrication system, pneumatic system components and ignition system for faulty / degraded conditions. The proposed methodologies have online capability and require only a limited set of microphones inside the APU compartment.